Dexterity's impact on critical hits

[Masamune]

o yea true strike lol.... k got my idiot moment for posting shit @ 5 A.M.

{Please forgive me}

[Yugl]

I know Ballista tends to skew DMG, but I can't imagine that it would affect WS Critical hit rates (The damage from them maybe, but I doubt the actual critical hit % changes). If possible, you could have two people set up and go to town. The only problem is that it requires two people (At least one with the bot program + Parse) and cannot be left overnight (Since you will time out eventually).

[Shamaya]

/comfort it's ok you can't always be perfect. What's more, WSC adds to your base damage so... Gonna hit for more than 0. Passed out early last night, gonna do dDex 50 prolly, while I'm out today.

[Masamune]

Well guess im stuck to continue with same bow+arrows on bunnies method, after this is done of course.

[CDF]

By the way, since the level 72 Ul'hpemde evasion value appears to be are correct (you could use /check; I haven't yet) and conversion of combat skill to accuracy is correct (no reason to believe it's been changed, right?), hit rate appears to take only integer percentages since my predicted hit rate with the formula is 88.5% and I obtained an estimate that is significantly less than 88.5%.

[Masamune]

at which error% that estimation were ?

and are you sure it was a lvl72 ? (not a lvl70 or 71?)

btw new data for you with dDEX21, ill post also my hitrate and accuracy relevant stats, cuze its only parse with lower than 95% accu i have.

regarding current crit%, KParser displays : 1207/10175=11.86% ...still going...
(for info i have also 11.06%=165/1491 from JPdata@dDEX20)
Maybe found 3rd jump?

[Byrthnoth]

Is 88.0% in the 95% confidence interval?

Until you test with exactly 1 more Acc and get 89% you won't know for sure. It could be that the Acc formula is just 1 Acc off. I doubt it has ever been tested this precisely before. Or one of the pieces of gear you're using could have a messed up description.
I'm not really trying to explain this away and would be unsurprised if it was the case, but I'm just listing the remaining things to be eliminated.

[Shamaya]

Tangent but for fighting bunnies you could try stage1 relic weapon along with the /sam and ranged weapon. So assuming the bunny wasn't killed (prolly still gonna die) you could possibly get off a ranged attack at the same time. For thf/sam, you could generate 40tp+ per mob doing that. As for CDF's tangent on hit rate, that's really cool bro. You should follow with Byrthnoth's suggestion probably.

[CDF]

Yes, 88% is in the confidence interval.

To reiterate, 307 - 280 => 13.5% net hit rate, so the overall predicted hit rate is 88.5%, but we observe for level 72 Ul'hpemde that the hit rate is significantly less than 88.5%, so this implies the the 1 point of accuracy from 306 to 307 appears not to do anything.

Surely, I could add one more accuracy and see if I get to 89%, as would be predicted, but that would merely be a confirmatory result rather than a discovery result, so to speak.

[Masamune]

For CDF, dDEX21 hit rate data:

Accuracy relevant stats used:
BLU75/sch1
Sword Skill=285
DEX=51+37=88
Gear accuracy+14(swiftbelt+EnkiduCap+IotaRing)
Accuracy Bonus JobTrait spellset
lvl71 Hpemde Evasion=275 (i'll double check after parse done)
BrutalEarring 5% DoubleAttack

Currently @ 10150/980=91.195% M.acc%

I don't know if that can help, since i didnot follow your accuracy testing, tell me if need more data?

Sidenote: anyone knows where i can find intimidations parsing results in KParser ? (im using BloodFeet)

[Masamune]

For Shamaya(and anyone wanting to test crit% Upper cap), i calculated the minimum needed sample size to obtain 0.5% error :

if cap=24%, N>28028
if cap=25%, N>28812

and while at it, i calculated also for other % (for other dDEX tests in region 43-48 ):
17% -> N>21683
18% -> N>22681
19% -> N>23649
20% -> N>24587
21% -> N>25493
22% -> N>26370
23% -> N>27230

.... after seeing those numbers, i wished i could assist in that region ;; GL for those who gonna attack those dDEXs. Will certainly need several nights in a row for a single dDEX test :/

[pchan]

dDEX=13 : 978/9454=10.34% +/- 0.61%
I'll continue it tonight but it seems to be 10% crit, which means the second tier happens at dDEX=14, although I doublt it happens every 7 dex since dDEX=39 gave 13% IIRC.

More sample size for the dDEX=13

I now totalize

dDEX=13 : 1978/19558=10.11% +/- 0.42% => 10%
I will probably do another 10k for dDEX=14 next unless I find gear to go in the dDEX=20+ range.

[Shamaya]

http://img.photobucket.com/albums/v7...te50dDex01.png
dDex50. *shrug* I'll work on making this larger tonight. It's not at all consistent with the 0.83.. bit. Might be cap, might not. I remember Searain's colibri data being outside 95% confidence.

edit: also Masa, assuming that the relationship is linear for 40+dex, we do not need that size of samples for each point. I wish I could explain it good but I can't. However pages ago, CDF explained it and made an example chart with confidence bands. Remember? Either way though I see what you're saying, and your reference numbers you just posted do help greatly :3. For the critical points, I feel happier (for some reason) getting a sample within 0.5% error. So I'll try to do that with dDex50 at least. So I'll continue the parse tonight and tomorrow.

[Masamune]

What's max dDEX you can achieve Pchan, me 22 is max, i definitely can't check the 4th jump predicted @ dDEX=26->27 (and maybe the 5th if there is one?).

i'll make an update tomorrow with new data posted.
So far, Crit% = floor(dDEX/7)+9% matches the 3 1st jumps we found.

Regarding my current parsing, i'll finish dDEX21, and launch dDEX20 this night.

I don't know what's current status of CDF ?

i know what you meant, but that 's where my thought diverges with CDF: while he and you think "safe enough" to assume region dDEX[40;50] is linear, i definitely don't. Even though crit% values @ dDEX=39 to 42 may lean towards the linearity asumption, nothing tell us that for example that crit% @ dDEX=45 will be in that line, at all.
Also, if yu trace a straight line passing through points dDEX39 40 41 and 42, you will see it points directly @ dDEX50=25%(or dDEX49=24%), while the very few reports we had before (including Searain's) were leaning towards crit% capping before, @ like dDEX=47ish. (invalid point forget it check EDIT below)
And last thing, my own parses for pDIF caps, though relatively low samples each, all added together for 16kish samples gives ~25.5% crit%.
That's why i prefer to think "i don't know" and not assume anything and test, hard way. Or a compromise would do like you started: starting points 40-42(done), then last ones (50-48 ), to finish with at least ONE point @ either 45or46 to prove the linearity.

But nonetheless, kudo for your great job so far

EDIT: checked Searain's data, all averaged together i got:
dDEX48=21.47%
dDEX50=23.03%
dDEX51=23.75%

[pchan]

range : animator+1 (4 dex)
head : pup AF if i request it (3 dex)
neck : 4 if I borrw ancient torque from mdk
ears : pilferer's earring if i go /thf (2), could get delta earring i guess with my rotting 150k ampulas (2 dex)
rings : iota(3) spinnel(4)
hands : elvaan rse (4)
body : -8 cos i need the -str from mahatma
back : AF2 (5 dex)
belt : rse (4 dex)
legs : rse (2)
feet : got to equip rse for -3 str

can get +29 without much trouble I guess, and maybe 32 if i borrow dex5 rings.

[Byrthnoth]

With only 8500 hits there's still a pretty large 95% confidence interval. Don't get disheartened yet!

Also, there's potentially a plateau in the 40-50 Dex range, if it's actually something like Floor(dDex*.833-26.65).

It just occurred to me that you killed 16,000 level 0 mobs for your pDIF caps Masa. Wowsas. That'd dedication.

[Byrthnoth]

I only hit submit once, but my connection is gay so I guess it sent twice.

[Shamaya]

I agree.. he is a freedom fighter, for a frenchy, haha.

An indication though that his pDif cap testing flew by underrated on these forums. (didn't make it pretty enough)

[Masamune]

Well... what can i say /blush XD

Yes my pDIF seem to not have been taken into consideration : i still see poeple around making dmg calculations with old pDIF formulas < < (except the new max 3.15 which has been already known)

But honestly, i could care less. On contrary, my RagingRush testing (and by extension all my research for other crit based WSs tests) failed terribly @ getting help. (Here i should have listened @ cdf "lolAlla" lol...). I dunno to be honest what to do for this matter, but sure i will never finish it alone.

Btw updated a draft @ wiki page for "Critical Hit Rate" discussion page.

[Motenten]

Aside: it would really be interesting to find evidence of random number generator failure (which would be really "LOL" since that's the driving force behind any MMORPG), but kparser can't record hits in order.

It can and does, but there's no current plugin to display that data. If you give me the specifics of what you want to know, I can write something up for you.

Sidenote: anyone knows where i can find intimidations parsing results in KParser ? (im using BloodFeet)

That should be on the Defense page, under Passive Defenses if you mean you intimidating the mob. If you mean the mob intimidating you... Hmm. I don't think I actually show that. I can look into finding a spot to put that if you're interested.


In addition, I can gear my sam to be able to get pretty comprehensive coverage of anything in the +20 to +36 dDex range, and up to +40 with a little finagling/borrowed gear, while still keeping str at 60. I'd be willing to help out with the testing if there's any particular points in that range that you want checked.